jbesw
commented
9 months ago Thanks - this will be live at https://serverlessland.com/repos/serverless-sentiment-analysis shortly.
Closed sachinh closed 9 months ago
jbesw
commented
9 months ago Thanks - this will be live at https://serverlessland.com/repos/serverless-sentiment-analysis shortly.
sachinh
commented
9 months ago Thanks for the review and approval.
Description
The purpose of this serverless pattern is to provide an automated, scalable, and cost-effective solution for processing audio recordings, transcribing spoken words into text, and analyzing the sentiment of the transcription. This pattern is particularly suited for organizations looking to automate customer service feedback analysis, call center operations, or any scenario where insights from audio recordings are critical, without the efforts of maintaining infrastructures. This solution includes:
Audio Transcription: Audio files recorded and uploaded to an S3 bucket trigger an AWS Step Functions state machine. The state machine coordinates with AWS Transcribe to convert the audio into a written transcript, making the content easily searchable and analyzable.
Sentiment Analysis: Once transcribed, AWS Comprehend is used to analyze the text for sentiment, determining if the overall sentiment is positive, negative, neutral, or mixed. This is essential for understanding customer feedback or the emotional tone in recorded interactions.
Notification and Escalation: The pattern includes mechanisms to notify relevant recipients of the sentiment analysis results. If the sentiment is negative, a notification can be sent via Amazon Simple Notification Service (SNS), and an email can be dispatched using AWS Simple Email Service (SES), allowing for immediate action or follow-up.
language
English
runtime
Python
Level
300
Type
Application
Use case
Interactive workload
Primary image
https://github.com/sachinh/serverless-sentiment-analysis/blob/main/images/architecture.png
IaC framework
AWS SAM
AWS Serverless services used
Description headline
Analyzing sentiment of audio recordings and trigger escalations using StepFunctions, SNS, and SES
Repo URL
https://github.com/sachinh/serverless-sentiment-analysis/
Additional resources
https://docs.aws.amazon.com/eventbridge/latest/userguide/eb-create-rule.html https://docs.aws.amazon.com/step-functions/latest/dg/concepts-input-output-filtering.html
Author Name
Sachin Holla
Author Image URL
No response
Author Bio
Solutions Architect @ Amazon Web Services. I am passionate about serverless programming, all things data, machine learning and developer automation.
Author Twitter handle
No response
Author LinkedIn URL
https://www.linkedin.com/in/sachinholla/
leave
No response